Variability in Aboriginal Identification between Census and Survey

aboriginal policy studies aps
Article
Who is Aboriginal? Variability in Aboriginal
Identification between the Census and the APS in 2006
and 2012
Claire Durand, Université de Montréal
Yves-Emmanuel Massé-François, Université de Montréal
Michael Smith, McGill University
Luis Patricio Pena Ibarra, Université de Montréal
aboriginal policy studies Vol. 6, no. 1, 2016, pp. 3-33
This article can be found at:

http://ejournals.library.ualberta.ca/index.php/aps/article/view/25410
ISSN: 1923-3299
Article DOI: https://doi.org/10.5663/aps.v6i1.25410
aboriginal policy studies is an online, peer-reviewed and multidisciplinary journal that publishes origi-
nal, scholarly, and policy-relevant research on issues relevant to Métis, non-status Indians and urban
Aboriginal people in Canada. For more information, please contact us at apsjournal@ualberta.ca or visit
our website at www.nativestudies.ualberta.ca/research/aboriginal-policy-studies-aps.
Who is Aboriginal? Variability in Aboriginal Identification
between the Census and the APS in 2006 and 2012
Claire Durand
Université de Montréal
Yves-Emmanuel Massé-François
Michael Smith
McGill University
Luis Patricio Pena Ibarra

Abstract: Over the last 50 years, censuses have shown very substantial increases in the
estimated sizes of Aboriginal populations in settler states such as Canada. Since these increases
cannot be explained by demographic factors alone, authors have proposed that, as the
situation of Aboriginal people has been under increasing public scrutiny, it has become more
socially acceptable to report that one is Aboriginal. This may be an explanation for increases
between censuses that are conducted five or ten years apart, but is such an explanation
plausible when comparing answers provided within six months of one another? This article
explores the factors associated with short-term fluidity in Aboriginal identification. In order
to do so, it uses Canadian data collected twice from among the same members of the defined
“population of Aboriginal identity” over a six-month period, in 2006 and in 2011–2012. Close
to a third of all Canadians who “identified” as Aboriginals in the Census long form or in the
National Household Survey (NHS) changed their answers when asked the same question in
the Aboriginal Peoples Survey (APS). Fluidity in identification depends on methodological
factors such as mode of administration and question wording. It also depends on individual
and contextual factors. Socioeconomic status and residence in an urban area or in specific
regions of Canada are the main factors that differentiate the three groups analyzed here—the
Fluid Indian/Métis, the New Métis and the New Indians—from the group that has a stable
identification. In light of this finding, we think that statistics produced on Aboriginal peoples
in Canada from the standard sources should be treated with some caution. Using the APS
identification numbers, for example, instead of those of the Census/NHS would likely reduce
the estimated differences between “non-Aboriginals” and “Aboriginals,” at least in terms of
education.
aboriginal policy studies, vol. 6, no. 1, 2016 3

www.nativestudies.ualberta.ca/research/aboriginal-policy-studies-aps
ISSN: 1923-3299
4 aboriginal policy studies
Introduction
Over the last 50 years, censuses have shown very substantial increases in the estimated sizes
of Aboriginal populations in settler states such as Canada (Goldmann and Siggner 1995;
Guimond, Robitaille and Sénécal 2007; Siggner and Costa 2005), Australia (O’Donnell
and Raymer 2015), New Zealand (Carter, Hayward, Blakely and Shaw 2009; Goodyear
2009), and the United States (Eschbach, Supple and Snipp 1998; Passel 1997; Snipp 1997).
Since these increases cannot be explained by demographic factors alone—births, deaths
and immigration—other explanations of such “explosions” in the count of Aboriginals
are proposed. Most of them revolve around a core argument: As the social, political,
and economic situation of Aboriginal people and groups has come under increasing
public scrutiny, it has become more socially acceptable to report that one is Aboriginal
(Andersen 2014a). This may be a plausible explanation for increases between censuses that
are typically conducted five or ten years apart, but is such an explanation relevant when
comparing answers provided within six months of one another? This article explores the
factors associated with the short-term fluidity in Aboriginal identification. In order to do
so, it uses Canadian data collected twice from among the same members of the defined
“population of Aboriginal identity” over a six-month period, in 2006 and in 2011–2012.
We first look at definitions and measurement of Aboriginality in order to place the
Canadian situation in context. We give specific attention to the Métis “category,” and
the ways in which different countries have dealt with intermixing, which is essential to
understanding fluidity in self-identification among Aboriginal people. Then we examine
the specific situation of Canada in order to understand why the fluidity in answers to
indicators of Aboriginality varies according to type of indicator, to categories of self-
identification, and to individual as well as contextual factors.
We conclude by discussing the possible consequences of this variability for research
using the Census, the NHS, and the APS, particularly as research examines the differences
between “Aboriginals” and “non-Aboriginals.”
Is There a Consensus Regarding Who Is Aboriginal?

International organizations encourage specific policies aimed at “Indigenous people,”
following the United Nations Declaration on the rights of Indigenous People in 2007.
Consequently, one would expect that there is a commonly used, agreed-upon definition of
the term. The UN, however, states:
Considering the diversity of indigenous peoples, an official definition of “indigenous”
has not been adopted by any UN-system body. Instead the system has developed
a modern understanding of this term based on the following: self-identification
as indigenous peoples at the individual level and acceptance by the community
as its member; historical continuity with pre-colonial and/or pre-settler societies;
a strong link to territories and surrounding natural resources; distinct social,
Who is Aboriginal? 5
economic or political systems; a distinct language, culture and beliefs; the forming
of non-dominant groups of society; and a resolve to maintain and reproduce their
ancestral environments and systems as distinctive peoples and communities.
(United Nations 2006)
If all the listed elements had to be present in order for someone to be classified as an
Indigenous person, this “modern understanding” of the term would be quite restrictive, or
at least ambiguous. It retains those Indigenous people who maintain a “traditional” way of
life and live in specific territories and it excludes those who live within “mainstream” society.
It emphasizes both a large, inclusive self-identification criterion, and a homogenizing,
stereotypical conception of Aboriginality that insists on the inferior economic and political
status of many Indigenous groups as an intrinsic feature of their condition. Many countries
have adopted definitions that pick up at least some elements of the UN definition (De
Costa 2014). Despite the fact that they are continuing to struggle to define Aboriginal, some
countries, including Canada (Guimond 1998), nonetheless try to quantify their Aboriginal
populations.
The Measurement of Aboriginality: Different Indicators

The UN definition hardly leads to a standard operational definition across countries, to say
nothing of how to deal with people of mixed origins and with the large share of Indigenous
people who live outside Indigenous communities. This definition can be translated via
two rather subjective indicators—the “form[ing of] non-dominant groups” and the
resolution “to maintain and reproduce their ancestral environments,” which cannot easily
be operationalized for purposes of counting—and five “objective” indicators referring to
self-identification, origin/ancestry, links to specific territories, language/culture, and legal
status.
Among the 21 countries in the Americas for which data are available—to which we add
Australia and New Zealand, two countries that share a common history of British colonial
rule with the US and Canada—there are thirteen countries that adopt self-identification in
their censuses. These questions take two forms: “Are you...” (four countries), and “Do you
consider yourself...” (eight countries). We could also classify among the self-identification
questions those that ask, “Do you belong to...” (twelve countries). The last form usually
refers to the notion of a group or nation.
Only five countries, including Canada, refer to origins/ancestry as an indicator of
Aboriginality. This indicator is closely related to the concept of “race”; it alludes to
“lineages” of descendants coming from a common ancestor (Antón, Bello, Del Popolo,
Paixao and Rangel 2009; Simon 1997). Seven countries refer to physical traits either in
question wording or in the categories themselves, while only four formally mention the
term “race”—including the United States until the 2010 Census.1
1 The United States Census Bureau announced in June 2015 that it would cease to use the term “race.”
As for territory of residence, four countries, including Canada, formally record

whether respondents live in an Indigenous community. Thirteen countries ask whether
respondents currently speak a native language or learned one in childhood. In Canada,
questions regarding Aboriginal languages are part of the Census long form/National
Household Survey. Three countries use the general term culture in the wording of the self-
identification question. In addition, five use questions on type of housing, and one uses
Indigenous spirituality as a category in the question on religion. Answers to these questions,
however, may be more useful for understanding the character of Aboriginal groups than
for counting them.
Finally, two countries refer to legal status. In Canada, the Indian Act specifies a legal
status for First Nations—that is, North American Indians, excluding people referred to as
Métis and Inuit—together with the conditions for entitlement to that status.
In short, there is no agreement on how Aboriginality should be measured and, more
importantly, to whom it refers. Self-identification, a subjective measure, is the most
widespread indicator across censuses, and also in research (Guimond, Robitaille and
Sénécal 2009). In addition, the boundaries of each type of measure are not as clear-cut as
it may seem. According to Stevens et al. (2015), when confronted with difficulties with the
existing categories, the countries that they studied, i.e. Canada, Australia and the United
States, tried to solve the problem by adding indicators. The US, for example, added a
question on Hispanic origins in 1980. Canada appears to be the country that uses the largest
number of indicators—origin, self-identification, status, and band membership. The two
last indicators are used to quantify the population of First Nations2 people entitled to live
in First Nation communities and to specific resources provided under the Indian Act, but
according to Goldmann and Siggner (1995) and Guimond, Kerr, and Beaujot (2004), these
population estimates differ substantially from the Indian Register.
The Measurement of Aboriginality: Categories and Boundaries
Measurement requires a quantification of attributes that have well-defined boundaries. For
example, Aboriginal Affairs and Northern Development Canada’s terminology states that
‘Indians, Métis and Inuit … are three separate peoples with unique heritages, languages,
cultural practices and spiritual beliefs (INAC 2012, emphasis added). But are they as mutually
exclusive as we tend to think?
Intermarriage has been widespread in the colonies since the beginning of the modern
colonial era, even more so in settler states where Europeans “took over” territories formerly
occupied by other people, i.e., in the Americas (Snipp 1997), Australia (Luke and Luke
1999), and New Zealand (Kukutai 2007). And intermarriage accelerates the blurring of
existing categories (Stevens et al. 2015) and the emergence of new ones (Stoler and Cooper
2 In Canada, the term First Nations is preferred to “North American Indians.” The term “Indian,” however,
appears in the Indian Act, which partly governs First Nations’ status and rights. It does so by establishing a
Register of Indians, which serves to determine the resources to which each band is entitled. Inuit, who live
mostly in Northern Canada, are not covered by the Indian Act.
2013). Being at the intersection of the legal and political fields (Simon 1997), censuses
have played a crucial role in the shaping of such categories through their influence on
decisions on how to measure, handle, include, or exclude those of “mixed race” (Andersen
2014b). Thus, the handling of mixed origins in censuses has varied over time and from
country to country (De Costa 2014; Yashar 2015) in reaction to international and national
sociopolitical contexts. Solutions have included providing a specific category or categories
for the intermixed (e.g., Métis/half-breed/Mestizo), allowing the selection of multiple
answers (e.g., White and North American Indian), or requiring respondents to select one
and only one category (Snipp 1997) on the basis of cut-off rules related to blood quantum or
to sex. This last option has become rare, since, as most censuses went to self-enumeration,
a growing number of respondents reported multiple ancestries or ethnicities, or refused to
choose just one.
In Latin America, six countries include a Métis category in a census question. However,
in Costa Rica, “Métis or White” is a single category; in Cuba, Mulattos are grouped with
Métis; and in Nicaragua, “Métis of the Caribbean Coast” is considered an Indigenous
group. According to Latino Barometro’s data for 2001 to 2013, close to 50 percent of the
total Latin American population identifies as Métis; that figure is higher than 60 percent
in many countries.
Other countries have opted for the multiple-response solution. In Australia, people can
declare up to two different ancestries (but not more than one ethnic identification). In New
Zealand, it is possible to declare membership in multiple ethnic groups. The US Census
started allowing multiple responses to the race question in 2000. People of mixed origins
could therefore select both White and Native American, and 40 percent of American
Indians did so, compared to five percent of Blacks, six percent of Hispanics, and fourteen
percent of Asians (Lavaud and Lestage 2005).
In Canada, categories aimed at counting those of mixed First Nations and European
origins appear in the censuses conducted in 1901, 1941, and continuously since 1981.
In 1901, census enumerators were instructed to “more precisely identify all origins of
the Métis” such as, for example, Ojibway French Breed (Hamilton 2007, 68). Gaps in
information between censuses are due to the absence of instructions to enumerators on
how to classify such people (Goldmann and Delic 2014; Goldmann and Siggner 1995).
According to Andersen (2008), in 1941, after forty years in which the Métis category had
disappeared from the Census, Métis had either become Indians, denied their background,
or chose to identify as French.
The Canadian Census started to record multiple origins in 1981. In 1996, the question
on ancestry/origins took its current form. It asks about the ethnic or cultural origins of
each person’s ancestors. A number of examples are provided, Aboriginal origins being
grouped together and included in the middle of the list according to frequency counts for
First Nations. Four write-in lines are provided, and up to six different origins are recorded.
Compared to Australia (where only two origins are recorded) and the United States,
the Canadian question and situation lead to the highest proportion of respondents with
multiple ancestry—thirty-five percent in 2006 (Stevens et al. 2015). Since 1986, the Métis
category has also been present as an Aboriginal category of its own in a self-identification
question. Therefore, with both a Métis category and the possibility of multiple origins,
Canada has opted for the most comprehensive solution to the quantification of Aboriginals
of mixed origins.
However, the self-identification question regarding Aboriginals, present in its current
form since 1996, requires those who identify as Aboriginals to skip the self-identification
question that is asked of other Canadians with categories like “white,” “black,” Chinese,
etc. Therefore, unlike US residents, Canadians cannot be both North American Indian
and White. And unlike Canadians, Americans are not given the choice of being Métis/
half-blood. In summary, people with multiple origins may either identify with each of their
origins, choose one origin or another, or identify with their own mixed category.
Indicating multiple “races” is not necessarily the same thing as indicating a unique
“mixed race” (Goldstein and Morning 2000), and this is even more obvious in the case of
Canada. The gender-based provisions of the Indian Act of 1869 conferred full status on the
descendants of any male Status Indian, regardless of his wife’s status, while on the other
hand, any female Status Indian who “married out” lost her status, and could not transmit it
to her children. Therefore, most people currently identified as Métis come from the latter
situation, even though their “quantum of Indian blood” may be similar to that of many
“Pure Indians” (Giokas and Chartrand 2002). Consequently, even though the meaning of
the two categories—Indian and Métis—has crystallized and been adopted by Aboriginals
themselves (Lawrence 2003, 2004; Sawchuk 2001; Simpson 2009), the terms still carry
inherent ambiguities due to their origins (Andersen, 2014b).
In addition, the term “Métis” encompasses two broad meanings. In Canada, it refers to a
specific national identity (Frideres and Gadacz 2012) adopted by people of mixed European
and Aboriginal heritage—often Cree and French—at colonial settlements that developed
around the fur trade economy in the Great Lakes region in the early seventeenth century,
and later at the Red River Colony in what is today Manitoba (Andersen 2011; O’Toole
2013). Second, in French, the term Métis denotes any person of mixed origins (Chartrand
1991). In this meaning, it has no specific territorial or historical roots; it simply means
being of mixed Aboriginal and European descent (Frideres and Gadaczs 2012). Since both
usages refer to mixedness (Giokas and Chartrand 2002), territorial and historical factors
are the only valid and relevant avenue for differentiating the two groups.
So who should be counted as Aboriginal? Anyone with Indigenous ancestors? That is
the most inclusive criteria, adopted by New Zealand, for example, and it was the solution
adopted by Canada before 1981. However, if all people of mixed Indigenous and European
origin were counted as Métis (this assumes that all such people would be aware of their
heritage) (Guimond et al. 2009), the result would be a very large number of Métis. The
estimates would probably be similar to what is seen in Latin America (Yashar 2015), since
there is no reason to believe that similar levels of intermixing did not happen and are
not happening presently in Canada. Therefore, one may conclude that the number of
declared Métis—or people of mixed origin—in a given country or region depends not only
on the level of intermixing but even more on historical and social acceptance as well as
legal recognition of intermixing, and therefore on the inheritance of the denomination by
succeeding generations. This question is relevant since, as we see below, the Métis category
is responsible for a high proportion of the increase in the estimated Aboriginal population,
in Canada as elsewhere.
The Explosion of Aboriginal Populations since the 1970s
In many countries, the number of people declaring Aboriginal origins or identifying as
Aboriginals has increased tremendously in the past fifty years (DeCosta 2014; Guimond
1998). These increases greatly exceed what could be explained by demography—that is,
by births, deaths, and immigration. An explanation for this, common among researchers,
is that change in the sociopolitical environment has encouraged Aboriginals to be proud
of their origins; this in turn is associated with an increased tendency of people of multiple
origins to identify as Aboriginals. Aboriginal identification has also been encouraged by
legal interventions, including allocation of resources. In practice, censuses’ shift from in-
person to self-enumeration has also been associated with increases in identification.
Canada’s current method of quantifying the Aboriginal population is the result of a long
process in which existing measures were modified and new measures appeared and in which
categories themselves appeared and disappeared (Goldmann and Delic 2014; Goldmann
and Siggner 1995). It is an interesting case study of a quest to find an appropriate and
accepted way of quantifying the Aboriginal population, a quest that became particularly
important in the context of a number of consequential events: the recognition in the 1982
Constitution of three distinct Aboriginal peoples, and therefore the official inclusion of
Métis among Aboriginal people; the modification of the Indian Act in 1985 in order to
respect the newly adopted Charter of Rights and Freedom and its prescriptions regarding
equality of treatment for men and women (Bill C-31); a major court decision on Métis
rights (R. vs. Powley 2003); and a further modification of the Indian Act that came into
force on January 31, 2011, just before the Census started, to allow grandchildren of Indian
women who had lost their status to regain it (Bill C-3).
However, substantial increases in the estimates of the population of Aboriginal origin
had started from 1971, when the Census went from enumeration to self-report. As can
be expected, growth rates vary according to the type of indicator, and they are affected by
changes in question wordings. From 1971 on, the total population declaring Aboriginal
origins started to increase at an average annual rate of around seven percent, a rate that was
far higher than the overall increase in the general population, which was around one percent
per year (Goldmann and Delic 2014). From 1991 to 1996, however, when the question on
origins was changed to a completely open-ended question,3 there was almost no increase
in the number of persons declaring an Aboriginal origin. The self-identification question
was asked in 1986 for the first time, and since then, growth rates have been somewhat
3 Before 1996, the question on origins proposed a number of categories from among which to choose,
providing boxes to check. There were categories proposed for Aboriginals, usually North American Indian,
Métis, and Inuit, and up to three spaces for write-in answers.
lower than they were for origins, i.e., three to five times (3.38 to 5.58) those of the general
population, including in 1996 (Goldmann and Delic 2014).
Though lower than for self-identification, the growth rates for Indian status and band
membership have also been much higher than should be expected when taking into
account births, deaths, and immigration. However, part of that increase can be explained
by the legal interventions that occurred during that period. According to Goldmann and
Siggner (1995), following Bill C-31, 85,000 persons regained their Indian status. The total
population of declared Status Indians went from 365,000 in 1991 to 461,000 in 1996, an
increase of twenty-six percent. In the following years, the growth rate decreased steadily,
from 20.9 percent between 1996 and 2001 to just over two percent between 2006 and 2011.
Growth rates differ also according to each category (Caron-Malenfant et al. 2014).
After the self-identification question appeared with the three categories included in the
Constitution, the Métis category saw an increase of 120 percent in its declared population
over the next five years. Figure 1 shows that, until 2011, the Métis category was responsible
for the major part in the increase in Aboriginal self-declaration. For example, the Métis,
who comprised thirty-one percent of the self-identifying Aboriginals in 2001, constituted
thirty-four percent of the population in 2006, because their group accounted for fifty
percent of the overall increase of twenty percent. And according to Andersen (2008), the
“Métis-as-mixed” meaning has played an important role in the demographic explosion of
the Métis category in Canada over the preceding three decades. This is attested to by the
fact that the explosion in the Métis self-declarations happened in regions outside the cradle
of the Métis nation, like New Brunswick and Nova Scotia (Andersen 2010). Conversely, the
growth rate for the Inuit population is much lower. However, this may be at least partly due
to the fact that Census data is collected in person by Statistics Canada’s staff in the remote
communities, which include all the northern communities in which the Inuit population
is concentrated.
Since 1996, the “measurement” of Aboriginality in Canada has stabilized in its current
form, with the same four questions asked. Only the examples provided in the question
on ancestors have been updated. The numbers continue to increase steadily—by amounts
greatly in excess of what could be expected from normal demographic increase (Caron-
Malenfant et al. 2014). However, this increase varies across Canada. It is systematically
lower in the northern territories, British Columbia, and the Prairies. Eastern Canada—the
Maritimes, Quebec and Ontario—stands out. This is where the growth rates are highest,
with some variation according to sub-region.
FIGURE 1. Increase in Aboriginal self-identification in the Census according to category,

1986–2011
Ethnic Mobility, Fluidity, and Categorical Mobility

The previous section illustrated the huge increase in Aboriginal populations in Canada, an
increase that is generally higher than can be explained by demographic factors. The only
possible explanation for the non-demographic substantial increase from one Census to the
next is that some respondents change their Census answers over time and those who report
an Aboriginal affiliation but did not do so before are more numerous than those who did
the reverse. These shifts have been called ethnic mobility (Guimond 2003; Siggner and
Costa 2005) and, sometimes, fluidity (Caron-Malenfant et al. 2014; Goldmann and Siggner
1995; Saperstein and Penner 2012). The term ethnic mobility assumes that the categories
that are presented to respondents have clear boundaries and are meaningful to them, but
this is not necessarily the case (Andersen 2014a; Lavaud and Lestage 2005). Andersen
(2014a) proposes the term categorical mobility instead, using the ambiguity of census
categories to explain the shift. Ambiguity is likely to be a particular problem for the term
Métis. In this article, we use the term fluidity (Caron-Malenfant et al. 2014; Goldmann and
Siggner 1995; Harris and Sim 2002; Saperstein and Penner 2012) to describe the variation
in answers given by the same individuals. This term does not presume any explanation of
the phenomenon. We also speak of identification rather than identity in order to convey the
fact that the information provided by respondents is an expressed identity, which may be
influenced by both the context and the questions used to elicit a response (Harris and Sim
2002).
The most fruitful way to study variability in answers is to have repeated responses
from the same individuals, a type of information that is rarely available. In Canada, Caron-
Malenfant et al. (2014) managed to match part of the Census public data file (20 percent
sample) for Census respondents from 2001 and 2006. In Australia, the Australian Bureau
of Statistics (ABS) produced a matched file for the period 2006–2011 for five percent
of Census respondents (Biddle and Crawford 2015; O’Donnell and Raymer 2015). This
allows for analysis of the mobility process over a five-year period. A few other authors
use longitudinal or panel data—not a census. Harris and Sim (2002) compared answers
to the question about race in the US by the same young individuals in different contexts
over a short period; Carter et al. (2009) compared answers to a question about ethnicity
over a three-year period in New Zealand; and Saperstein and Penner (2012) analyzed two
instances of self-identification by the same individuals in 1979 and 2002, and a time-series
of 20 instances of identification by an interviewer between 1979 and 1998 in the US.
What influences the probability of fluidity? Previous research suggests that, not
surprisingly, respondents with more than one origin are more likely to switch their
identification (Caron-Malenfant et al. 2014; Harris and Sim 2002). In Canada, being
legally recognized as an Indian reduces the likelihood of a switch out of Aboriginal self-
identification and increases the likelihood of a switch into the category (Caron-Malenfant
et al. 2014; Goldmann and Delic 2014).
It is also clear that responses are influenced by the categories and options presented to
respondents. In its main question on Aboriginal status, the ABS severely limits options. It
allows three categories: Aboriginal, Torres Strait Islander, or neither. That question does
not allow multiple identifications, though another question—on ancestries—allows two
options. New Zealand lists five indigenous groups. Canada provides three major categories—
First Nations, Inuit, and Métis. Most research uses data from the whole population, and
shows that multiple responses and variability in answers are both higher for Aboriginals
than for any other group (Biddle and Crawford 2015; Caron-Malenfant et al. 2014; Carter
et al. 2009; Harris and Sim 2002; Saperstein and Penner 2012).
We chose self-identification as the focus of our study of variability for a number of
reasons. It is the indicator that the UN considers the most appropriate and the most
widespread as a measure of ethnicity in the Americas. It is the indicator that is the most
frequently—indeed, almost uniquely—used in academic and governmental research
(Andersen 2013). It is an indicator whose wording has not changed over time and, unlike
origins, it is present in both APS surveys. Finally, unlike status or band membership, it
includes all Aboriginals, not just First Nations.
Four types of factors may explain fluidity in identification. First, respondents’ individual
characteristics, such as age, sex, and spouse’s origins, may be associated with fluidity. In
Canada, since the Indian Act had gender-based provisions and women who married out
before 1985 lost their status, it is conceivable that women may be more inclined to change
their declarations. Indeed, Caron-Malenfant et al. (2014) show that changing from non-
Aboriginal to Aboriginal is more likely among women. However, the findings of others
(Biddle and Crawford 2015; Carter et al. 2009) show no relationship to sex. Age reflects
individual characteristics as well as cohort/period effects related to the changes that have
occurred over the last fifty years. Research show that fluidity over five years is higher among
younger respondents (Biddle and Crawford 2015; Caron-Malenfant et al. 2014; Carter et
al. 2009). Similarly, marrying a non-Aboriginal person is also associated with more fluidity
(Biddle and Crawford 2015).
The socioeconomic status of individuals is an important factor in predicting change in
identification. Saperstein and Penner (2012), for example, show that there is a tendency for
those who have improved their socioeconomic status to choose to be white —“whitening”
or “passing”—when they can do so. Conversely, research shows that those who change
their identification from non-Aboriginal to Aboriginal are less likely to have a higher
education and income (Biddle and Crawford 2015; Caron-Malenfant et al. 2014; Saperstein
and Penner 2012).
Generally, the more ethnically heterogeneous the place of residence, the more fluid
the identification (Biddle and Crawford 2015). For example, living in an urban area is
generally associated with greater variability (Biddle and Crawford 2015), but with lower
odds of switching to Aboriginal identification (Caron-Malenfant et al. 2014). In addition,
fluidity is associated with certain regions of the country due to their historical and social
characteristics (Biddle and Crawford 2015; Caron-Malenfant et al. 2014; Harris and Sim
2001; Saperstein and Penner 2012). Caron-Malenfant et al. (2014) show that, for Canada, the
odds of changing one’s self-identification to North American Indian are higher in Ontario
than everywhere else in Canada, and the odds of revealing a Métis identification are higher
in the Prairies and British Columbia and lower in Eastern Canada than in Ontario.
This article aims first to identify the factors associated with variations in the fluidity of
answers according to indicator and to categories. Second, it aims to identify the individual
and contextual factors that are associated specifically with variability in self-identification.
Following the literature review, we expect origins and self-identification to be the most
fluid indicators, and we expect Métis and First Nation to be the most fluid categories of
self-identification. However, living—and consequently having lived—in a First Nation
community should be associated with stability among First Nations. As for fluidity in self-
identification, it should be higher among women and younger respondents. Having a non-
Aboriginal spouse is known to be related to fluidity; however, this variable was not available
in the data file. A high socioeconomic status should translate to less change in the direction
of an Aboriginal identification. As well, residents of urban areas should be less likely to
switch to Aboriginal identification. Finally, changing specifically to a Métis identification
should be more frequent in the Prairies, and changing to North American Indian should
be more frequent in Ontario.
Methodology
The Aboriginal Peoples Survey (APS) is a post-censal survey administered six months after
the Census. The sampling frame for the survey is provided by the Census. Answers to the
APS can therefore be linked to those given by the same individuals in the Census. This
provides an unusual opportunity to study fluidity. Because similar questions are asked of
the same persons (conditional on them belonging to the “Aboriginal identity population”)
twice in a six-month interval, broader changes in the sociopolitical environment are
unlikely to explain observed fluidity, and any relevant changes would be easily identifiable.
For this research, Statistics Canada gave us access to the responses provided by APS
respondents to the four questions tapping Aboriginality that they had asked in the Census/
NHS. This provided us with panel data that allowed us to study variability in identification
at the individual level. Unlike the data used by Caron-Malenfant et al. (2014) and Biddle
and Crawford (2015), these are not files in which records are matched as closely as possible
according to the similarity of answers to a number of variables, and provided that people
did not change address, but instead, here, individual data are linked; and contrary to Harris
and Sim (2002), Carter et al. (2009), and Saperstein and Penner (2012), the data are both
census and post-census data. However, and contrary to all these authors, the data are
restricted to Aboriginals. Finally, contrary to other researchers, we look at change among
Aboriginal categories and not just between non-Aboriginals and Aboriginals; as well,
contrary to Caron-Malenfant et al. (2014), we include Inuit.
Data
The compulsory Census long-form questionnaire was sent to one in five households in
2006. In 2011 the Canadian government decided to discontinue the compulsory Census
long form. Statistics Canada replaced it with the National Household Survey (NHS),
administered to one in three households. The response rate for the NHS was deemed
“acceptable”: 77 percent overall, ranging from 70 percent in Prince Edward Island to 84
percent in the Northwest Territories (where it was conducted by interviewers). These
“long-form” questionnaires were given to every household in First Nation communities
(that is, on reserves) and to Inuit communities.
Anyone who declared or was reported as having an Aboriginal origin or identification,
or as being a Status or Treaty Indian or a band member in the 2006 Census long form or
in the 2011 National Household Survey, was categorized as a member of the Aboriginal
population. This was the sample frame for the APS, conducted six months later. It was
administered to one in five members of the Aboriginal population, but excluded those
living on First Nations communities, often called reserves. Statistics Canada derived two
concepts (Statistics Canada 2012) for the purpose of deciding who should answer the APS:
the population of Aboriginal origin only—those who declared Aboriginal origins but no
identification, status, or band membership; and the population of Aboriginal identity,
which encompassed all those who either declared identification as Aboriginal or had legal
First Nation (FN) status or membership in a band.
The 2006 APS was administered to those who belonged to the Aboriginal population.
In 2012, however, respondents had to belong to the population of Aboriginal identity in
the APS in order to be eligible. Therefore, those who had declared an Aboriginal identity
in the NHS but not in the APS were dropped in 2012. In order for both surveys to be
comparable, we excluded all those who did not identify as Aboriginals in the 2006 APS.4
This reduced sample is the comparable 2006 sample. Given our interest in self-identification,
we restricted our analysis to respondents 15 years and over.5 There were originally 29,520
respondents fifteen years and over to the 2006 APS, representing a total of 1,038,390 people.
Excluding those who did not identify as Aboriginals in the 2006 APS6 leaves us with 24,660
respondents, representing 749,200 persons. There were 20,720 respondents of the same age
group to the 2012 APS, representing 795,180 persons. The analyses were performed on the
comparable samples.
In interpreting the results, we must bear in mind the following. First, the census form
is completed by a “primary respondent” who answers for other members of the household,
while the APS is completed by the selected individuals themselves.7 This means that the same
person may not have answered both surveys. In 2006, 44.2 percent of the APS comparable
sample respondents were Census primary respondents; and in 2012, this figure was 48.2
percent. Second, the Census/NHS is self-administered, while the APS is conducted by an
interviewer over the telephone for the most part. However, in remote communities, both
surveys are administered in person by an interviewer. These methodological features may
explain some of the variability. Being contacted to participate specifically in the Aboriginal
People Survey may induce greater identification as Aboriginal. Conversely, the fact that
both the Census/NHS and the APS are conducted in the same way in remote communities,
i.e., in person by an interviewer, may lead to lower variability among residents of these
communities. Third, a few communities in Yukon and in the Northwest Territories8 were
excluded in 2012 but not in 2006. Therefore, the two surveys sample slightly different
populations. Fourth, values were imputed for 1.3 percent of weighted respondents who
did not identify in one or the other group of Aboriginals in 2012.9 In 0.8 percent of the
4 Those who declared that they were status Indians or members of a band but did not identify as Aboriginals
were kept in the 2012 APS but not in the comparable APS 2006 sample. These respondents represent a tiny
proportion of the population (in 2012, 1.3 percent). Our decision was justified by the fact that, in 2012, part
of the answers to the self-identification question were imputed using the answers to the questions on status
and band membership (see text).
5 APS is also conducted among five-to-fourteen-year-olds. The questions are answered by a parent or by the
adult in charge in the young person’s household.
6 They account for 16.5 percent of the unweighted sample and 27.8 percent of the weighted sample. Most of
them did not identify as Aboriginal in both surveys. Those who did identify as aboriginal in the Census but
not in the APS comprise 4.6 percent of the whole sample.
7 The APS is answered by a proxy for half of the fifteen-to-seventeen-year-olds, and exceptionally for adults
who cannot answer for health-related reasons.
8 In 2012, these communities were excluded in order not to interfere with another survey that was conducted
at about the same time (Statistics Canada 2012).
9 Information provided by Statistics Canada staff upon request.

cases, imputation used the NHS answer. This can only reduce variability, but the impact
is negligible. In the other cases (0.5 percent), those who reported a legal status or band
membership but did not identify as Aboriginals have been reassigned to the FN category
of self-identification. This prevents us from having strictly comparable samples, but the
possible impact is negligible.
Finally, since the only data available are those pertaining to Aboriginals living outside
of First Nation communities, the results represent most declared Métis and Inuit people,
but only about half of First Nations. The proportion of Status Indians living outside First
Nations communities varies considerably by region, nation, and band. In 2011, it ranged
from 28 percent in Quebec to 97.9 percent in the Northwest Territories (Amorevieta-Gentil,
Bourdeau and Robitaille 2015). It is around forty percent in the Atlantic Provinces and the
Prairies, around fifty percent in Alberta and British Columbia, and sixty-three percent in
Ontario. Since residents of First Nations communities have to be Status Indians and are
more likely to have a stable identification (Goldmann and Delic 2014; Caron-Malenfant et
al. 2014), this exclusion is likely to increase the overall figures for fluidity and may have an
impact on the differences according to place of residence.
Measures
The Measurement of Aboriginality and of Fluidity

The 2006 Census long form and the 2011 National Household Survey use the same
four indicators to categorize Aboriginal people. The first question pertains to origins.
Respondents are asked: “What were the ethnic or cultural origins of this person’s ancestors?”
The following then clarifies that “An ancestor is usually more distant than a grandparent.”
Respondents are provided with the following list of examples in 2006: Canadian, English,
French, Chinese, Italian, German, Scottish, East Indian, Irish, Cree, Mi’kmaq (Micmac),
Métis, Inuit (Eskimo), Ukrainian, Dutch…, etc. In 2012, the list included “Salish” after
“Mi’kmaq.” The position of the Aboriginal sub-categories is determined by the frequency
count of the total population of First Nations in the preceding Census. All the Aboriginal
categories are grouped together with, for First Nations, the names of the most frequently
mentioned Nations. In remote communities, however, the list of examples is different, and
Aboriginal categories are at the beginning of the list. The positioning of categories on the
list may have consequences for the answers given to the question. There was a “Count
me Canadian” campaign in 1991 which led to four percent of respondents writing in
“Canadian.” Consequently, the category “Canadian” was put fifth on the list of examples in
1996. The proportion selecting that category that year jumped to twenty percent (Stevens
et al. 2015, 28).
The second question asks whether the respondent considers him/herself—or the person
for whom he/she is answering—Aboriginal. Respondents may select one or more of the three
proposed categories: North American Indian, Métis, or Inuk (Eskimo). When respondents
identify as or are reported to be Aboriginal, as stated earlier, they are asked to skip over the
question on self-identification that is asked of other Canadians. Therefore, those of mixed

origins may either identify as Métis, combined or not with another Aboriginal category,
or they can choose to identify according to only one of their origins—i.e. First Nation or
“White”—but not both.
Two more questions are asked of all Canadians, including those who did not identify
as Aboriginals: whether the person is a Status Indian and whether he/she is a member of
an Indian band (and which band). These two questions are highly correlated. In 2006, only
2.1 percent of Status Indians were not members of a specific band, and in 2012, 3.9 percent.
In the 2006 APS, the question about origins directly asked whether the respondent had
ancestors who belonged to an Aboriginal group. The respondent could select one or more
of the three categories already mentioned. In the 2012 APS, this question was not asked.
The self-identification question, asked in both surveys, differs from the census question
mostly in the order of the answers. In the census, no comes before yes, while in the APS,
it is the opposite. As well, in 2012, the question was divided into two separate questions,
one relative to identifying as an Aboriginal person and the other relative to the category or
categories with which the person identified.
Finally, the questions pertaining to legal status and band membership are the same
as in the Census except that the name of the band is not requested. The two measures are
somewhat less correlated in the APS than in the Census. In 2006, five percent of those who
reported being Status Indian in the APS did not report being band members, and in 2012,
5.4 percent.
In order to answer the research question regarding fluidity in self-identification,
we computed a new variable that allowed for differentiating the type of fluidity in the
declarations. We grouped the respondents who vary their declarations into three types.
1. The fluid Indian-Métis: Those who identified, or who were identified, with a
different category in the APS than in the census/NHS (Indian/First Nation in one
and Métis in the other or vice-versa) or who added or subtracted a category (Indian
in one and Indian and Métis in the other or the reverse). This category includes
twenty-five different combinations, and involves mostly respondents who shift
between the Indian/First Nation and Métis categories. They make up 10.3 percent
of the sample in 2006 and 8.9 percent in 2012.
2. The new Métis: Those who did not report—or for whom the primary respondent
did not report—an Aboriginal identification in the census but who did identify as
either only Métis—in the majority—or as Métis and Indian or Inuit. These people
“come out,” i.e., they reveal themselves as mixed. They make up 9.2 percent of the
respondents in 2006 and 8.3 percent in 2012.
3. The New Indians: Those who did not report—or for whom the primary respondent
did not report—an Aboriginal identification in the census but identified as Indian/
First Nation or Inuk (a much smaller group) in the APS. They make up 12.5 percent
of the 2006 respondents and 9.7 percent of the 2012 respondents.10
10 The New Métis and New Indians are essentially people who are part of the population of Aboriginal origin
only, according to their answers to the Census questions. However, since they identified as Aboriginals in the
APS, they were included in the sample.
The fourth group comprises those for whom exactly the same identification was reported
in both surveys; they comprise 68 percent of the sample in 2006 and 71 percent in 2012.
Factors Related to Fluidity in Identification
Following the literature review, four main groups of factors are included in the analyses.
The two first groups of factors are at the individual level. The first group comprises sex
and age group. There are three age groups: fifteen to thirty-four, thirty-five to fifty-four,
and fifty-five and over. The second group of factors comprises indicators of socioeconomic
status, i.e., years of schooling in four categories (less than high school, high school, college,
and university) and quartiles of income. The third group is related to the context in which
people live. The first indicator is the Metropolitan Influence Zone (MIZ), which comprises
four ordinal categories of urbanness, from the Census Metropolitan Area itself (CMA)
to “none or weak.” The second is region, which comprises the usual regions of Canada,
i.e., the Maritimes, Quebec, Ontario and the Prairies. However, we grouped the northern
territories, including Yukon and Nunavut, together with British Columbia in order to have
sufficient variation in terms of the Aboriginal groups and places of residence and large
enough sample sizes in each category. This latter grouping serves as a reference category to
which all the other regions are compared.
Finally, we added being primary respondent in the census in both the 2006 and 2011–
2012 analyses in order to control whether some of the variability could be due to the fact
the same person did not necessarily answer both surveys.
We took advantage of the questions asked in the 2012 APS to add two variables related
to life experience: having previously lived on a First Nation community, and having
attended a residential school. We do not include other indicators of Aboriginality among
the predictors of variability since these factors are quite endogenous and therefore produce
very high and possibly unstable coefficients in a multivariate regression analysis.
Analysis
In order to examine variability according to indicators, descriptive data and tables of the
relationship between self-identification and the other indicators are presented. In order to
examine the factors associated with variability in identification, we conducted multinomial
logistic regressions. The stable group is the reference category to which the three types of
fluidity are compared. The predictors were entered hierarchically in order to check for
possible mediation effects, i.e., substantial variations in the coefficients upon entering new
predictors. Since entering the blocks hierarchically did not reveal any major mediation
effect, the results presented are those of the final analyses.
Results
Variability in Indicators
Table 1 presents first the number of Aboriginals estimated by the APS and the estimates
computed from APS respondents’ answers to the Census.11 It shows that for the indicators
that are present in both cycles, the APS estimates are always higher than the Census/NHS
estimates.12 As expected, the larger increase over the six-month period comes from the self-
identification question; it shows a net increase of 27.8 percent in 2006 and 21.9 percent in
2012. The increase is high, in the range of thirteen to eighteen percent, even for indicators
that are deemed “objective” such as legal status and band membership. In 2006, the increase
for origins is close to fourteen percent. Part of this increase may be attributed to the effect
of switching from an open-ended question, in which one may feel free to write in or not
his/her Aboriginal origins, to a direct, closed one, in which the respondent is asked whether
he/she has Aboriginal origins—yes or no.
Table 1. Comparison between APS and Census answers according to indicators
2006 (comparable sample) 2011-2012
CENSUS APS Increase Variability NHS APS Increase Variability
Aboriginal Origins 650800 741330 13.9% 40.9% 694200 Not Asked N/ A N/ A

Aboriginal Self-
586430 749200 27.8% 32.0% 652300 795180 21.9% 27.0%
Identification
Legal Indian Status 242890 287000 18.2% 10.7% 263150 297140 12.9% 10.1%
Band Membership 239230 277830 16.1% 9.8% 252730 294690 16.6% 10.6%
Unweighted N 24660 20720
Weighted N 749200 795180
How do these increases compare with intra-individual variability? For the 2006
comparable sample, 40.9 percent had a different answer for origins between the Census
and the APS: 32 percent changed their identification, 10.7 percent changed their answers
regarding whether they were Status Indians, and 9.8 percent changed their answers regarding
band membership. In 2012, the situation was quite similar, with 26.9 percent varying their
response to the self-identification question, 10.1 percent varying their response to the
12 The APS is weighted in order to match the estimates for the whole Census in terms of origins in 2006 and
of population of aboriginal identity in 2012. The figures presented are different from Census figures because
of the variability in answers (Statistics Canada, 2012).
13 These estimates are based on answers to the Census by APS respondents. The respective Census estimates
for self-identification, for example, are 631,280 for 2006 and 699,750 for 2012 according to the complete
Census data, not far from the APS-Census estimates i.e., 634,140 and 652,300.
status question, and 8.3 percent varying their response to the band membership question.
Therefore, as expected, fluidity is higher for self-identification and for origins, though
information on origins is only available from 2006. Although fluidity is undoubtedly
associated with the increase in estimates, the relationship between the two is not similar
for the different indicators for a number of reasons, including methodological ones. The
higher fluidity in the origins and self-identification questions is related to the fact that, for
these indicators, there are three main categories of Aboriginality, while status and band
membership are yes-or-no questions.
Table 2. Fluidity in self-identification according to origins
ORIGINS - CENSUS ORIGINS - APS ORIGINS - NHS
Type of fluidity Indian Métis Inuit Indian Métis Inuit Indian Métis Inuit
Fluid Indian/ Métis 57.5% 33.2% 2.5% 79.5% 60.7% 6.4% 56.8% 26.3%
New Métis 51.8% 44.9% 0.8% 51.6% 92.0% 2.0% 56.5% 42.2% Estimates
too small for
New Indians 84.1% 6.7% 1.9% 94.9% 7.6% 3.4% 89.0% 4.9% publication
Stable 53.0% 30.3% 6.0% 64.0% 47.5% 7.0% 53.0% 70.3%
Table 2 shows the relationship between variability in self-identification and declared

origins. If we look at the Census figures for 2006, the profile of the Fluid Indian/Métis
is close to the profile of stable respondents and not much different from that of the New
Métis. A little more than half declared a First Nation origin, and close to a third declared
Métis origin. However, if we compare these numbers with similar numbers in 2012, they
are similar for the Fluid Indian-Métis and the New Métis groups, but are quite different for
the stable group. This group is much more likely to have declared Métis origin in 2012 (70.3
percent) than in 2006 (30.3 percent). In addition, the report of different Aboriginal origins
in the 2006 APS is much higher than in the Census for most categories. Among the New
Métis, for example, the Métis origin more than doubles, from 45 percent in the Census to
92 percent in the APS, for the same individuals.
Table 3. Fluidity in self-identification according to legal status and band membership
Legal Status Band Membership Legal Status Band Membership
Census APS Census APS Census APS Census APS
Fluid Indian/ Métis 33.7% 39.3% 32.8% 37.2% 28.1% 31.3% 25.2% 30.7%
New Métis 1.5% 8.0% 1.0% 6.8% 1.2% 4.0% 0.6% 4.5%
New Indians 6.8% 19.4% 8.1% 22.2% 6.2% 13.9% 5.6% 17.8%
Stable 41.0% 46.0% 46.4% 43.8% 41.0% 45.0% 39.6% 44.1%

As expected, one of the major distinctions among the different types of fluidity is in
the proportion of people with First Nation legal status or band membership. As shown in
Table 3, in 2006, people in the Fluid Métis-Indian group are almost as likely as the stable
respondents to have legal status or to be band members, i.e., between thirty percent and
forty percent of the Fluid Métis-Indian group had legal status or were band members
compared to forty to forty-six percent of stable respondents.
However, in 2012, the proportion is somewhat lower among the Fluid Indian-Métis,
i.e. between twenty-five and thirty-one percent. These same proportions go down to one
percent in the Census and to four to twenty percent in the APS for the New Métis; they
reach five to eight percent in the Census and fourteen to twenty-two percent in the APS for
the New Indian group. Conversely, more than forty percent of the stable group is either a
Status Indian or a band member.
These results suggest three conclusions. First, as expected, legal status and band
membership are associated with more stability in identification. However, these statuses are
also associated with fluidity among aboriginal categories, a new finding from this analysis.
This information is in line with the literature that stressed the fuzzy boundaries between
the Indian and Métis categories (Andersen 2014b; Lawrence 2004). Second, APS data show
more coherence than Census data. Respondents align their identification with their origins
and status. Third, we have to keep in mind that there are substantially more people of Métis
origin in the stable group in 2012 than in 2006 because this difference in composition may
have an influence on the factors that differentiate stable respondents from the others.
Variability in Categories of Self-Identification
Who are these respondents who report a shift in identification across Aboriginal categories,
or who switch from no identification to an identification as Aboriginal? Table 3 presents
the results of the multinomial regressions conducted on the 2006 and 2012 APS data.
Overall, the Nagelkerke pseudo R2, which allows for comparison of the models with and
without predictors, gives a value of 13.1 percent in 2006 and 10 percent in 2012.13 Adding,
in 2012, the variables for having lived on a First Nation community and having attended a
residential school increases the Nagelkerke pseudo R2 to 13.2 percent.
The Fluid Métis-Indian Group
The odds14 of being in this fluid group rather than in the stable group are higher for men
than for women. This runs counter to our hypothesis, which stated that either women
would be more likely to be fluid or there would be no difference according to sex. Since
13 The Nagelkerke pseudo R2 is a revised version of Cox and Snell pseudo R2, adjusted so that the maximum
value is 1 (Tabachnick and Fidell 2012, 462). Hosmer and Lemeshow (2013, 185) notice, however, that pseudo
R2values obtained in logistic regression are generally low compared to R2 values in linear regression.
14 The odds ratio, presented here, is “the change in odds in being in one of the categories of the dependent
variable when the value of a predictor increases by one unit” (Tabachnick and Fidell 2012, 463).
previous researchers did not examine movements within the Aboriginal group separately,
this finding is new. The odds of being in the fluid group rather than in the stable group are
also higher for those over 55 years old than those in the fifteen-to-thirty-four age group
(significant in 2006 only). This also runs counter to our hypothesis stating that fluidity is
expected to be higher among younger people. Regarding socioeconomic status, the odds
of being in the Fluid Indian-Métis group rather than in the stable group are lower if one
is better-educated, and, according to the 2012 APS, lower also when respondents benefit
from an income higher than the median. Since previous research largely neglected this
group, we did not have specific hypotheses.
Table 4. Factors related to fluidity in self-identification
2006 2011-2012
Reference category: stable declarations Fluid Métis/Indian New Métis New Indian Fluid Métis/Indian New Métis New Indian
EXP(B) EXP(B) EXP(B) EXP(B) EXP(B) EXP(B)
Intercept (B) - 2.537 *** - 3.358 *** - 3.255 *** - 2.240 *** - 2.968 *** - 3.074 ***
Individual Characteristics
Being a man 1.356 *** .936 1.013 1.125 * 1.083 1.037
Age (Ref= 15 - 34 years)
35-54 years .982 .980 .988 .884 .917 1.189 ***
55 and more 1.340 *** 1.284 *** 1.085 1.134 1.312 *** 1.243 ***
Socio-economic Characteristics
Schooling (Ref=less than high school )
High school diploma 1.034 1.511 *** 1.285 *** .955 1.350 *** 1.099
College diploma .878 * 1.216 ** .957 .911 1.128 1.092
University diploma .803 * 1.283 ** 1.017 .663 *** 1.403 *** 1.324 ***
Income (Ref=below the 25th percentile )
Between 25th & 50th percentile 1.041 .885 1.189 * .865 .969 .941
50th to 75th percentile .882 1.177 * 1.160 * .748 *** .942 .957
75th percentile and over .897 1.548 *** 1.369 *** .659 *** .947 .783 ***
1
Urbanity (Ref= MIZ - weak & none)
MIZ Strong & Moderate 1.391 *** 2.283 *** 2.540 *** 1.447 *** 1.412 *** 2.661 ***
CA (Census Agglomeration) 1.205 ** 1.844 *** 2.361 *** 1.325 *** 1.233 * 1.890 ***
CMA (Census Metropolitan Area) 1.305 *** 2.655 *** 3.220 *** 1.622 *** 1.410 *** 2.661 ***
Region (Ref=Northern Canada & BC )
Prairies 1.421 *** .909 .506 *** .857 * 1.500 *** .446 ***
Ontario 1.321 *** 1.203 * 2.108 *** 1.237 *** 1.745 *** 2.019 ***
Quebec 2.931 *** 3.688 *** 4.512 *** 2.427 *** 4.362 *** 3.640 ***
Provinces of the Atlantic 1.371 ** 2.263 *** 6.637 *** 1.333 *** 1.764 *** 4.067 ***
First Nation experience
Has lived in a FN community .623 *** .190 *** .170 ***
Has attended a boarding school .925 .444 *** .290 ***
Was first respondant in Census .972 1.100 * .937 .999 1.001 1.048
Statistics
N (rounded & unweighted) 24660 20720
Pseudo R-squared (Nagelkerke) .131 .132
Thresholds: *=0.05, **=0.01 and ***=0.001
1
Metropolitan Influenced zone
The results show that the odds of being in the Fluid Indian-Métis category are higher
if one lives in any zone of urban influence rather than in zones of weak or no urban
influence—that is, in remote or isolated communities. Since both Métis and First Nations—
and presumably those who shift between the two—are historically less-present in the north,
where most remote communities are located, and since Inuit, a very stable group, are more
present in this area, this result is expected.
The odds of being in the fluid group are higher for those living in Eastern Canada—
in Ontario, in the Maritimes and especially in Quebec—than in British Columbia or
northern Canada. So at first glance, this group may be more likely to include Métis-as-
mixed respondents rather than “Métis of the Red River”. However, for the Prairies, the
results of 2012 and 2006 go in opposite directions. In 2006, the odds of being in the Fluid
Indian-Métis group were forty percent higher (1.42) when living in the Prairies rather than
in BC/Northern Canada, while in 2012 they were close to seventeen percent lower (.857).
Descriptive statistics (not presented here) confirm these results. The fluid group accounts
for 11.6 percent of Aboriginals in the Prairies in 2006, but only 7.7 percent in 2012. This
may be explained by the differences in the composition of the stable group between 2006
and 2012. All the other coefficients, however, are quite similar across the two cycles. Events
that occurred between 2006 and 2102 (which we have not identified) may explain this
situation.
Finally, the fact of the APS respondent being a primary respondent—or not—in the
Census has no impact on the odds of being in this category. For 2012, as expected, the
odds of being in the fluid group rather than stable are 1.6 times lower (.623) among those
who have lived in a First Nation community before. However, having attended a residential
school is not a significant factor in predicting the odds of being among the fluid category.
Entering the Métis Category or Multiple Categories: The New Métis
The New Métis and the stable group have a similar distribution of men and women. The
literature on this question was contradictory, but Caron-Malenfant et al. (2014) found that
“becoming” Aboriginal was more common among women. In the context of this study,
these results are not confirmed. The odds of being a New Métis rather than remaining
stable are higher in the 55-and-over age group than in the younger group in both cycles.
This also runs contrary to our hypothesis, based on the literature, which shows higher
fluidity among younger people.
The odds that respondents will be in the New Métis category increase among all those
who have a diploma—high school, university, or anything in between. This is also the case
for those who have an income above the median according to the 2006 data. However,
income does not contribute significantly to explaining the odds of entering this category in
2012. This goes against our hypothesis, which was based on previous research.
The odds of being a New Métis also rise when respondents are living in or close to
cities rather than in rural/remote communities. This confirms our hypothesis: the more
heterogeneous the place of residence, the more fluid the identification. Respondents in
this group were more likely to live in Ontario, the Maritimes, and particularly Quebec (3.7
times more likely in 2006 and 4.3 times more likely in 2012) than in British Columbia or
northern Canada. This further confirms that the New Métis are more likely to be Métis as
mixed. These results do not confirm the hypothesis based on the results presented by Caron-
Malenfant et al. (2014). As is the case for the Fluid Indian-Métis, there is an inconsistency
between the 2006 and 2012 results for the Prairies: the 2012 data show that the odds that
Prairie residents will be in this group increase 1.5 times compared with residents of BC and
the north, while the coefficient for 2006 is not significant. What happens with this group
is almost the mirror of what happens with the Fluid Indian-Métis group. Since both have
to do with mixedness, there may be some fuzzy boundaries between the two groups; the
results may also be explained by compositional differences between 2006 and 2012, or by
events that occurred between 2006 and 2012 that may have influenced the situation and
identification of the Métis in the Prairies.
Finally, the 2012 data show that having lived in a First Nation community decreases
substantially the odds of being a New Métis (an odds ratio of .19, which means 5.3 times
lower than those who never lived in a First Nation community). The odds are 2.2 times
lower for those who attended a residential school. Since the policy on residential schools
applied mostly to Status Indians and Inuit living in remote communities, this is consistent
with expectations.
In 2006 only, the odds of being in this group are slightly higher when the APS had been
completed by someone who was a primary respondent in the Census.
Entering the Indian or Inuk categories: The New Indians
The odds of entering the New Indians group differ between 2006 and 2012 in terms of
individual characteristics. As with the preceding group, sex is not related to entering the
category. There is no difference by age when using the 2006 data, but, according to the
2012 data, the odds are higher for older respondents, i.e., those who are older than 35 years.
Again, the literature shows that entering the Aboriginal group is more likely among the
young people. So our results for 2012 run contrary to those of other researchers.
The 2006 data show that the odds of entering the New Indians group increase with a
high school diploma compared to no diploma, but do not differ with a higher diploma,
while the 2012 data show that the odds increase only with a university degree. While the
2006 data show that the odds of being a New Indian increase for respondents who are in
the higher percentiles of income, the 2012 data show the reverse. As for the New Métis,
the results do not confirm our hypothesis regarding education. However, the 2012 data
confirms our hypothesis regarding income, contrary to the 2006 data.
The 2006 and 2012 data present the same picture for the impact of place of residence.
They reveal that the odds of entering the New Indian group compared to being stable
are clearly higher for those who live in urban areas. They are also higher for those who
live in Ontario (twice as high), Quebec (4.5 times and 3.6 times as high respectively) and
especially the Maritimes (6.6 times and 4.1 times as high). This confirms Caron-Malenfant
et al.’s (2014) results for Ontario, but not for the other regions. In addition, the odds are
twice as low among those who live in the Prairies than among those who live in BC and in
northern Canada. Our results are, for most regions, quite different from those of Caron-
Malenfant et al. (2014).
Finally, the 2012 data show that the odds of being a New Indian rather than remaining
stable are much lower among those who have ever lived in a First Nation community (.17)
and who have attended a residential school (.29). In short, living in FN communities and
having attended residential schools are associated with a stable identification. Those who
were primary respondents in the Census do not differ from those who were not.
Summary
The main blocks of variables—age and sex, education and income, urban residence,
region—all contribute significantly to the explanation of fluidity. However, the individual-
level variables explain no more than 1.8 percent of the variance according to Nagelkerke
pseudo R2, which is very low. The Metropolitan Influence Zone explains 1.8 percent of
the variance in 2006 and one percent in 2012. The main contributor to the explanation is
region, which contributes 9.2 percent to the total variance in 2006 and 7.7 percent in 2012.
The second-largest contributor to the explanation, present only in 2012, is First Nation
experience. It accounts for 3.2 percent of the variation. Finally, fluidity was not higher in
instances in which the APS was answered by someone other than the Census’s primary
respondent.
The results of the effects of individual characteristics reported here differ from
what previous research has found. This may be because there are major methodological
differences between our study and those of both Caron-Malenfant et al. (2014) and Biddle
and Crawford (2015). These authors compared reported Aboriginal status over the five-
year period between Censuses. The Censuses were conducted in similar ways with the
same questionnaire. In contrast, we have two different modes of survey administration
and somewhat different question wordings. Our design is better compared to Harris
and Sim (2002), who showed that identification varies with the mode of administration
(self-administered vs. in-person), the presence or absence of others when the in-person
interview is conducted, and the question asked—particularly whether or not it is possible
to report belonging to multiple “races.”
One way to make sense of our results and reconcile them with Caron-Malenfant et
al. (2014) and Biddle and Crawford (2015) is to consider that, for example, the higher the
socioeconomic status (i.e. greater education, income, etc.), the higher the tendency not to
declare an Aboriginal identification in the Census. The New Métis and the New Indians in
our sample do indeed have higher education level and, in 2006, higher income. Our results
point to the fact that, faced with the forced choice to be either Aboriginal or white, and in
the absence of social pressure as a result of the fact that the Census was self-administered,
a number of Aboriginals—apparently those with the higher socio-economic status—in
conformity with the literature (Biddle and Crawford 2015; Caron-Malenfant et al. 2014;
Harris and Sim 2002; Saperstein and Penner 2012) chose the “white” option. In the APS,
their Aboriginal identification is activated by the context of the survey, and they are not
offered the possibility of being “white”; they can only choose not to identify as Aboriginals
and therefore, in 2012, be dropped from the survey.
It is also possible that we tap into something different from other authors: the complexity
of identification and, especially, the reality of fuzzy boundaries among categories, particularly
for people of mixed origins, along with the problems associated with the “forced choice”
in the Census self-administered questionnaire. It is, in our view, likely that a substantial
proportion of Aboriginals feel that they can choose in which circumstances they identify
as Aboriginals.
Caron-Malenfant et al. (2014) and Biddle and Crawford (2015) attribute the change in
identification that they observe to sociopolitical changes between the two Censuses. We
also note differences in the results for 2006 and 2012, particularly in the variation in the
presence of the Fluid Indian/Métis and the New Métis groups in the Prairies. This may also
reflect sociopolitical change in the situation related to the Métis as a Nation in that region
during that period. The composition of the stable group in 2012 tends to confirm that the
people with a stable identification were much more likely to be of Métis origin in 2012 than
in 2006.
Isolating a Fluid Indian-Group separate from those who switch from non-Aboriginal
to Aboriginal allows for a first insight into the fluidity of identification among Indians,
possibly of multiple origins. This group shows different characteristics from the two other
fluid groups, in that they are more likely male, have less formal education, and are more
likely to reside in Quebec than in other regions. This fluid group was, in a way, the missing
block in the study of the consequences of intermixing for identification. It shows that the
blurring of the boundaries is ubiquitous. However, the impact of a Fluid Indian/Métis group
on estimates of the population may be neutral because similar numbers are going from
each category to the other (Caron-Malenfant et al. 2014).
The substantial fluidity in the answers to the two different questions on origins in 2006
hint at problems with the Census write-in open-ended question and its examples. This
question is very imprecise: it does not give any time-frame within which ancestry should
be considered, and the examples that are given may induce people to choose among the
categories or feel less involved when their own Nation is not listed.
The study presented here is not perfect. We compare answers to questions that are asked,
sometimes in different ways, and that are sometimes answered by different respondents,
using measures aimed at the same concept but asked in a different context. This may have
a number of consequences, which we have tried to point out and control for to the fullest
extent possible. Nevertheless, some of the results that emerged directly contradict what
is generally concluded in the literature. We tried to make sense of these differences, but
further analysis would perhaps allow for a more in-depth understanding of the situation. It
would be appropriate to replicate the results in the next rounds of the Census and the APS,
or using previous rounds.
Conclusion
In view of the UN’s “modern understanding” of who is Aboriginal; in the absence of a
definition of who should be included in this population and why; and given the wording
of the questions asked, it is not surprising that some people provide different answers in
different contexts. Counting Aboriginals requires a delineation of boundaries, but those
boundaries will inevitably be blurred. There are the standard methodological decisions
to be made on survey presentation, question wording, and allowed response categories
discussed above. We think it is impossible to come up with a single right resolution to these
methodological issues, though some decisions are certainly better than others.
At least as important is the character of the Aboriginal population itself. After centuries
of intermixing with settler populations, people’s self-identification as Aboriginal will very
often involve choice. That choice is, in turn, influenced by the various factors identified
in our model. The growth in the number of self-identified Aboriginals shows that choices
change over time. All this suggests that the use of Aboriginal counts for both policy design
and general research should always be accompanied by some discussion of both likely
methodological effects and social determinants of the numbers a count reveals.
This paper has shown that close to a third of all Canadians who “identified” as
Aboriginals in the Census long form or in the National Household Survey changed their
answers to the self-identification question when asked the same question in the Aboriginal
People’s Survey. The shifts in reported Aboriginal status are not random; they are related to
some important characteristics of the relevant population. In light of this finding, we think
that statistics produced on Aboriginal peoples in Canada from the standard sources should
be treated with some caution. Using the APS identification, for example, instead of that of
the Census/NHS would likely reduce the estimated difference between “non-Aboriginals”
and “Aboriginals,” at least in terms of levels of formal education.
The variability in responses we have identified suggests that Statistics Canada might
usefully undertake an in-depth inquiry into the phenomenon. Comparisons between
Aboriginals and other Canadians are complicated by the fact that the self-identification
questions designed by Statistics Canada are different for the two groups. It is unfortunate
that it is too late to implement changes to the 2016 Census that would allow those of
Aboriginal origins to elect as many identifications as they want, like any other Canadians.
A long, shared history means that a very large number of Canadians can be considered,
to some degree or another, Métis. What we see in these data is most probably a prelude to
what will be seen for many other settlers who have come to Canada (or elsewhere) more
recently. As such, more and more Canadians will be expected to have multiple affiliations,
and we will need to decide how to ask questions on self-identification that are meaningful
for respondents as well as useful for social statistics.
Bibliography
Amorevieta-Gentil, M., R. Bourbeau and N. Robitaille. 2015. «Les mouvements migratoires
des Premières Nations: Reflets des inégalités.» Population Change and Lifecourse
Strategic Knowledge Cluster Discussion Paper Series/Un Réseau stratégique de
connaissances Changements de population et parcours de vie Document de travail
3(2): 2.
Andersen, C. 2008. “From Nation to Population: The Racialisation of “Métis” in the

Canadian Census.” Nations and Nationalism 14(2): 347–68. doi:10.1111/j.1469-
8129.2008.00331.x.
Andersen, C. 2010. “Mixed ancenstry or Métis?” In Indigenous Identity and Resistance:

Researching the Diversity of Knowledge, edited by B. Hokowhitu, N. Kermoal, and
C. Andersen, 23–36. Portland: Otago University Press.
Andersen, C. 2011. “MoyaTipimsook (‘The People Who Aren’t Their Own Bosses’):
Racialization and the Misrecognition of ‘Métis’ in Upper Great Lakes Ethnohistory.”
Ethnohistory 58(1): 37–63. doi: 10.1215/00141801-2010-063.
Andersen, C. 2013. “Underdeveloped Identities: The Misrecognition of Aboriginality in

the Canadian Census.” Economy and Society 42(4): 626–50. doi:10.1080/03085147.
2012.760350.
Andersen, C. 2014a. “Ethnic or Categorical Mobility? Challenging Conventional

Demographic Explanations of Méis Population Growth.” In Aboriginal Populations:
Social, Demographic, and Epidemiological Perspectives, edited by F. Trovato and A.
Romaniuk, 263–84. Edmonton: University of Alberta Press.
Andersen, C. 2014b. Métis: Race, Recognition, and the Struggle for Indigenous Peoplehood.
Vancouver: UBC Press.
Antón, J., Á. Bello, F. Del Popolo, M. Paixão, and M. Rangel. 2009. Afrodescendientes en
América Latina y el Caribe: del reconocimiento estadístico a la realización de derechos.
CEPAL: Serie Poblacion y desarrollo N.87. Santiago de Chile.
Biddle, N., and H. Crawford. 2015. “The Changing Aboriginal and Torres Strait Islander
Population: Evidence from the 2006–11 Australian Census Longitudinal Dataset.”
In CAEPR Indigenous Population Project 2011 Census Papers, No.18, 36 p.
Caron-Malenfant, É., S. Coulombe, E. Guimond, C. Grondin, and A. Lebel. 2014. «La

mobilité ethnique des Autochtones du Canada entre les recensements de 2001 et
2006.» Population 69(1): 29–55. doi:10.3917/popu.1401.0029.
Carter, K.N., M. Hayward, T. Blakely, and C. Shaw. 2009. “How Much and for Whom
Does Self-Identified Ethnicity Change over Time in New Zealand? Results from a
Longitudinal Study.” Social Policy Journal of New Zealand 36: 32–45.
Chartrand, P. 1991. “Aboriginal Rights: The Great Métis Dispossession.” Osgoode Hall Law
Journal 29(3): 457–82.
De Costa, R. 2014. “Descent, Culture, and Self-Determination: States and the Definition
of Indigenous Peoples.” Aboriginal Policy Studies 3(3): 55–85. doi: 10.5663/aps.
v3i3.22227.
Eschbach, K., K. Supple, and C.M. Snipp. 1998. “Changes in Racial Identification and the
Educational Attainment of American Indians, 1970–1990.” Demography 35(1): 35–
43. doi:10.2307/3004025.
Frideres, J.S., and R.R. Gadacz. 2012. Aboriginal Peoples in Canada. Ninth Edition. Toronto:
Pearson.
Giokas, J., and P.L. Chartrand. 2002. “Who Are the Métis in Section 35? A Review of the
Law and Policy Relating to Métis and ‘Mixed-Blood’ People.” In Who Are Canada’s
Aboriginal Peoples? Recognition, Definition and Jurisdiction, edited by P.L. Chartrand,
83–125. Saskatoon: Purich Publishing Ltd.
Goldmann, G., and S. Delic. 2014. “Counting Aboriginal Peoples in Canada.” In Aboriginal
Populations: Social, Demographic, and Epidemiological Perspectives, edited by F.
Trovato and A. Romaniuk, 59–78. Edmonton: University of Alberta Press.
Goldmann, G., and A. Siggner. 1995. “Statistical Concepts of Aboriginal People

and Factors Affecting the Counts in the Census and the Aboriginal Peoples
Survey” (paper presented to the 1995 Symposium of the Federation
of Canadian Demographers, Ottawa, October 23-25, 1995).
Goldstein, J.R., and A.J. Morning. 2000. “The Multiple-Race Population of the United
States: Issues and Estimates.” Proceedings of the National Academy of Sciences 97(11):
6230–35. doi:10.1073/pnas.100086897.
Goodyear, R.K. 2009. The Differences Within: Diversity in Age Structure between and within
Ethnic Groups. Wellington: Statistics New Zealand.
Guimond, E. 1998. “Ethnic Mobility and the Demographic Growth of Canada’s Aboriginal
Populations from 1986 to 1996.” In Report on the Demographic Situation in Canada,
1998-1999, 187-200. Statistics Canada: Ottawa. Catalogue no.91-209-XIE
Guimond, E. 2003. “Changing Ethnicity: The Concept of Ethnic Drifters.” In Aboriginal

Conditions: Research Foundations for Public Policy, edited by J. White, P. Maxim and
D. Beavon, 91–107. Vancouver: UBC Press.
Guimond, E., D. Kerr, and R. Beaujot. 2004. “Charting the Growth of Canada’s Aboriginal
Populations: Problems, Options and Implications.” Canadian Studies in Population
31(1): 55–82.
Guimond, E., N. Robitaille, and S. Sénécal. 2007. «Définitions floues et explosion

démographique chez les populations autochtones du Canada de 1986 à 2001» (paper
presented in Montréal at Statistiques sociales et diversité ethnique conference
co-organized by the Centre interuniversitaire québécois de statistiques sociales
[CIQSS] and the Institut national d’études démographiques [INED], Montreal,
December 6-8, 2007).
Guimond, E., N. Robitaille, and S. Senécal. 2009. «Les Autochtones du Canada: une
population aux multiples définitions.» Cahiers québécois de démographie 38(2):
221–51. doi:10.7202/044815ar.
Hamilton, M.A. 2007. “‘Anyone Not On the List Might as Well be Dead’: Aboriginal
Peoples and the Censuses of Canada, 1851–1916.” Journal of the Canadian
Historical Association/Revue de la Société historique du Canada 18(1): 57–79.
doi:10.7202/018254ar.
Harris, D.R., and J.J. Sim. 2002. “Who is Multiracial? Assessing the Complexity of Lived
Race.” American Sociological Review 67(4): 614–27. doi:10.2307/3088948.
Hosmer, D.W., Jr. and S. Lemeshow. 2013. Applied Logistic Regression. Third Edition. New
Jersey: John Wiley & Sons.
Indigenous and Northern Affairs Canada (INAC). 2012-10-01. Terminology. http://www.

aadnc-aandc.gc.ca/eng/1100100014642/1100100014643. Accessed May 2015.
Kukutai, T.H. 2007. “White Mothers, Brown Children: Ethnic Identification of Maori‐
European Children in New Zealand.” Journal of Marriage and Family 69(5): 1150–
61. doi:10.1111/j.1741-3737.2007.00438.x.
Latino Barometro. 2015. Latino Barometro, Opinion Publica Latino

Americana. http://www.latinobarometro.org/lat.jsp. Accessed May 2015.
Laura Stoler, A., and F. Cooper. 2013. Repenser le colonialisme. Paris: Payot.
Lavaud, J.-P., and F. Lestage. 2005. «Compter les indiens.» L’année sociologique 55(2): 487–
517. doi:10.3917/anso.052.0487.
Lawrence, B. 2003. “Gender, Race, and the Regulation of Native identity in Canada and the
United States: An Overview.” Hypatia 18(2): 3–31. doi:10.1353/hyp.2003.0031.
Lawrence, B. 2004. “’Real’ Indians and Others: Mixed-Blood Urban Native Peoples and
Indigenous Nationhood. Vancouver: UBCPress.
Luke, C., and A. Luke. 1999. “Theorizing Interracial Families and Hybrid Identity: An
Australian Perspective.” Educational Theory 49(2): 223–49. doi:10.1111/j.1741-
5446.1999.00223.x.
O’Donnell, J., and J. Raymer. 2015. “Identification Change and its Effect on Projections
of the Aboriginal and Torres Strait Islander Population in Australia.” Journal of
Population Research 32(3-4): 297–319. doi:10.1007/s12546-015-9155-7.
O’Toole, D. 2013. “From Entity to Identity to Nation: The Ethnogenesis of the

Wiisakodewininiwag (Bois-Brulé) Reconsidered.” In Métis in Canada: History,
Identity, Law and Politics, edited by C. Adams, G. Dahl, and I. Peach. 143–203.
Edmonton: University of Alberta Press.
Passel, J.S. 1997. “The Growing American Indian Population, 1960–1990: Beyond
Demography.” Population Research and Policy Review 16(1-2): 11–31.
doi:10.1023/a:1005724610787.
Saperstein, A., and A.M. Penner. 2012. “Racial Fluidity and Inequality in the United States.”
American Journal of Sociology 118(3): 676–727. doi:10.1086/667722.
Sawchuk, J. 2001. “Negotiating an identity: Métis political organizations, the Canadian

Government, and Competing Concepts of Aboriginality.” The American Indian
Quarterly 25(1): 73–92. doi:10.1353/aiq.2001.0012.
Siggner, A.J., and R. Costa. 2005. “Trends and Conditions in Census Metropolitan Areas.”
In Aboriginal Conditions in Census Metropolitan Areas, 1981–2001: Ottawa, ON:
Statistics Canada. Catalogue no. 89-613-MIE- No.008, 40 p.
Simon, P. 1997. «La statistique des origines: L’ethnicité et la «race» dans les recensements
aux États-Unis, Canada et Grande-Bretagne.» Sociétés contemporaines 26(1): 11–44.
doi: 10.3406/socco.1997.1442.
Simpson, A. 2009. “Captivating Eunice: Membership, Colonialism, and Gendered

Citizenships of Grief.” Wicazo Sa Review 24(2): 105–29. doi:10.1353/wic.0.0031.
Snipp, C.M. 1997. “The Size and Distribution of the American Indian Population: Fertility,
Mortality, Migration, and Residence.” Population Research and Policy Review 16(1–
2): 61–93. doi:10.1023/a:1005784813513.
Statistics Canada. 2012. Aboriginal Peoples Survey. Detailed information for 2012.
http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=3250&l
ng=en&db=imdb&adm=8&dis=2#a2. Accessed February, 2016.
Stevens, G., H. Ishizawa, and D. Grbic. 2015. “Measuring Race and Ethnicity in the Censuses
of Australia, Canada, and the United States: Parallels and Paradoxes.” Canadian
Studies in Population 42(1–2): 13–34.
Tabachnick, B.G., and L.S. Fidell. 2013. Using Multivariate Statistics. Sixth Edition. Boston:
Pearson.
United Nations. 2006. “Permanent Forum on Indigenous Issues Fact Sheet: Who Are
Indigenous Peoples?”In Fifth Session, Special Theme: The Millennium Development
Goals and indigenous peoples: Re-defining the Millennium Development Goals, 2.
United Nations Headquarters, New York, 15 to 26 May 2006. http://www.un.org/
esa/socdev/unpfii/documents/5session_factsheet1.pdf. Accessed May 2015.
Yashar, D.J. 2015. “Does Race Matter in Latin America? How Racial and
Ethnic Identities Shape the Regions’s Politics.” Foreign Affairs,
https://www.foreignaffairs.com/articles/south-america/2015-02-16/does-race-
matter-latin-america. Accessed February 2016.
Acknowledgements:
The authors wish to thank, first, the three reviewers who did a tremendous job. Their very
detailed and relevant comments on the first version and the supplementary information
that they provided allowed us to improve the article greatly. We also wish to thank the
personnel at Centre Interuniversitaire Québécois de Statistiques Sociales (CIQSS), the
Research Data Center at Université de Montréal, for their vital help in getting the necessary
data and information to carry out this research. We extend our appreciation as well to
Statistics Canada personnel who helped us understand the peculiarities of the data and the
decisions that were taken to collect data the way it was collected. Finally, we wish to thank
SSHRC, grant no. 435-2013-1042 to Michael Smith and Claire Durand, for its financial
support for this research project.

Variability in Aboriginal Identification between Census and Survey

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Variability in Aboriginal Identification between Census and Survey

Uploaded by

Copyright:

Available Formats

aboriginal policy studies aps

aboriginal policy studies Vol. 6, no. 1, 2016, pp. 3-33

This article can be found at:

Luis Patricio Pena Ibarra

aboriginal policy studies, vol. 6, no. 1, 2016 3

Is There a Consensus Regarding Who Is Aboriginal?

The Measurement of Aboriginality: Different Indicators

As for territory of residence, four countries, including Canada, formally record

FIGURE 1. Increase in Aboriginal self-identification in the Census according to category,

Ethnic Mobility, Fluidity, and Categorical Mobility

9 Information provided by Statistics Canada staff upon request.

The Measurement of Aboriginality and of Fluidity

question on self-identification that is asked of other Canadians. Therefore, those of mixed

2006 (comparable sample) 2011-2012

CENSUS APS Increase Variability NHS APS Increase Variability

Aboriginal Origins 650800 741330 13.9% 40.9% 694200 Not Asked N/ A N/ A

2006 (comparable sample) 2011-2012

ORIGINS - CENSUS ORIGINS - APS ORIGINS - NHS

Table 2 shows the relationship between variability in self-identification and declared

2006 (comparable sample) 2011-2012

Legal Status Band Membership Legal Status Band Membership

Census APS Census APS Census APS Census APS

Stable 41.0% 46.0% 46.4% 43.8% 41.0% 45.0% 39.6% 44.1%

Andersen, C. 2008. “From Nation to Population: The Racialisation of “Métis” in the

Andersen, C. 2010. “Mixed ancenstry or Métis?” In Indigenous Identity and Resistance:

Andersen, C. 2013. “Underdeveloped Identities: The Misrecognition of Aboriginality in

Andersen, C. 2014a. “Ethnic or Categorical Mobility? Challenging Conventional

Caron-Malenfant, É., S. Coulombe, E. Guimond, C. Grondin, and A. Lebel. 2014. «La

Goldmann, G., and A. Siggner. 1995. “Statistical Concepts of Aboriginal People

Guimond, E. 2003. “Changing Ethnicity: The Concept of Ethnic Drifters.” In Aboriginal

Guimond, E., N. Robitaille, and S. Sénécal. 2007. «Définitions floues et explosion

Indigenous and Northern Affairs Canada (INAC). 2012-10-01. Terminology. http://www.

Latino Barometro. 2015. Latino Barometro, Opinion Publica Latino

O’Toole, D. 2013. “From Entity to Identity to Nation: The Ethnogenesis of the

Sawchuk, J. 2001. “Negotiating an identity: Métis political organizations, the Canadian

Simpson, A. 2009. “Captivating Eunice: Membership, Colonialism, and Gendered

You might also like